Say I am doing a time series prediction which predict some value for next time step with past T inputs from historical inputs. Say I am using a RNN module like LSTM or GRU.
In trainning/validation, I fed RNN module with batches of shape (batch_size, T, *) data to train a model.
When inferencing, I can either:
- Always use past T inputs to get next step prediction, then discard the state of the RNN module. That is: use input from time -T to -1 to get prediction at t=0 (last output of LSTM or GRU module), then discard the final state of the RNN module and use input from time -T+1 to 0 to get prediction at t=1 etc.
- keep the RNN state, and each time feed only one input to get the prediction. That is: first use input from time -T to -1 to get prediction at t=1 like above. Then keep the current state of the RNN and feed the RNN with input at t=1 to get the prediction at t=2 etc.
Which one is better? Or It depends on specific problems? Thanks